A server accepts and processes requests from clients. The server keeps the results of the most recent requests in memory as a cache. A new request arrives on average every 25 ms. The processing of each request takes 15 ms. If the requested result is not in the memory cache, additional 75 ms are needed to access the disk. On average, 80% of all requests can be serviced without disk access. The creation of a new thread takes 10 ms. Required:
a. Should the server be implemented as a single-threaded or a multi-threaded process?
b. What percentage of disk accesses would have to be satisfied without disk access for the two approaches to break even?